Why to record your voicebank under a blanket? Has boring Science!

    Share

    irei1as

    Posts : 265
    Notes : 1007
    Reputation : 40
    Join date : 2010-05-13

    UTAU Information
    UTAUloid or Project: Valen Sodebades
    Blog/Website: www.youtube.com/user/irei1as
    Status: Almost there... but kinda failed

    Why to record your voicebank under a blanket? Has boring Science!

    Post by irei1as on Sat Aug 14, 2010 6:40 pm

    (Everything of this may be false! XD
    So you can just ignore it.
    I was just bored and sometimes I like to search random things around the internet. That's why this is in the randomspamfuntiemz.)

    You are alone in your room and you try to record something with your mic and you find that the recording hasn't a perfect quality. You try again with a blanket around your head and the microphone and the recording is a lot better... why?

    If you say it's because external noise is in the way and the blanket stops it... you're kinda right! That external noise is, in part, your own echo and the blanket what does is to stop it for appearingappearing.

    In agreement with this article:
    "On the other hand, researchers have found that echoes in the range of 1 to 50 milliseconds are lumped together by the brain with the direct sound, so they are not actually heard as distinct echoes. Delays greater than 50 milliseconds are heard as echoes."

    -Note: wikipedia here states:
    "When dealing with audible frequencies, the human ear cannot distinguish an echo from the original sound if the delay is less than 1/10 of a second."
    0.05 s from that obscure article or 0.1 s from wikipedia... I've choosen the smaller because I have faith you've got super humans hearing.)


    That is, if your voice does an echo of under 50 ms your brain is going to ignore that sound and put it together with your normal voice. (Possible exception:
    human echolocation)

    Let's going to calculate how long takes that hypothetical echo.
    Let's say the situation is like this:



    Where "L" is the distance to the other wall of your room.

    Now, you know: velocity = space / time
    As we want time we work out the value of time:
    time = space / velocity

    space is 2·L because the sound goes to and has to come back from that wall.
    velocity is speed of sound.

    You can make your own numbers with your own L and speed of sound (as speed of sound changes with temperature -and material but I guess you like to sing with air-).
    A good aproximation for speed of sound is 343 meters per second (dry air and 20ºC = 68ºF).
    Let's say the wall of our room is at 4 meters, then:
    time = 2·4/343 = 0.023 second = 23 milliseconds

    That is lower than the 50 ms so that echo is ignored by our brain.

    For our brain to get that echo the room should be larger:
    space = velocity · time
    space = 343 · 0.050 = 17.15 meters
    But that space is 2·L so L is half that: L is a bit over 8.5m
    The room should be larger than 8.5 meters for us to hear the echo.

    But, why we hear echo with an empty room a lot smaller than that? That's because the sound can bounce a lot more in an empty room:

    (Note: only between two walls in the picture. There are more possible bounces and so more distance between the four walls).
    When the room is full of forniture the secondary bounces are going to die faster without coming back. Actually the recording rooms have special walls to absorb even the 1st bounce so you don't actually need that blanket... but I doubt we can record our "poor" voicebank in an isolation booth like Momo Momone's voicer did (I read that somewhere, not sure if it's true).

    So, as
    time = space / velocity
    but now space is something like 6·L (only 3 bounces) and that's 24 meters for our 4 meters room, higher than the 17.15m and we get a time over that 50 ms so our brain distinguishes the echo.


    So, ok, there is an echo in our room that humans can't hear. But why our computer records it?
    If we follow the Nyquist–Shannon sampling theorem a good sampling rate of our recordings is going to be 44.1 kHz (the common one for UTAU). That's 44100 Hz or 44100 samples of sound in one second. If there are 44100 samples in a second that means between each sample there is 1/44100 seconds, that is around 0.00002 seconds between each sample or around 0.02 milliseconds. That's a lot smaller than those 50 ms, the extra noise from the echo is going to be recorded.

    If you want that echo down, a blanket works because echo is more possible with a hard and smooth surface (like that wall or even that wood wardrobe).

    Using wall carpets is another way of reducing the echo outside isolation booths because the carpet, the same way the blanket does, is going to stop the bouncing of the sound.

    Darkkira1332

    Posts : 516
    Notes : 1072
    Reputation : 12
    Join date : 2010-05-15
    Age : 23
    Location : In My room, working and tweaking my utaloid

    UTAU Information
    UTAUloid or Project: CVUV series (2 characters created)
    Blog/Website: I don't have one yet.
    Status: Kira:Ver. 2 (creating act.1 VB), Lyrica( Releasing Act.1 VB)

    Re: Why to record your voicebank under a blanket? Has boring Science!

    Post by Darkkira1332 on Wed Aug 18, 2010 10:24 am

    thanks for the info!!!!

      Current date/time is Tue Dec 06, 2016 8:09 am